Speechless: Speech Instruction Training Without Speech for Low Resource Languages

Research paper by Alan Dao and collaborators. Published 2025-05-23. Topics: large language models, LLM, speech instruction data, TTS, semantic representation.

By Alan Dao, Dinh Bach Vu, Huy Hoang Ha, Tuan Le Duc Anh, Shreyas Gopal, Yue Heng Yeo, Warren Keng Hoong Low, Eng Siong Chng, Jia Qi YipResearch paper

Resource information

Authors: Alan Dao, Dinh Bach Vu, Huy Hoang Ha, Tuan Le Duc Anh, Shreyas Gopal, Yue Heng Yeo, Warren Keng Hoong Low, Eng Siong Chng, Jia Qi Yip
Research area: Language technology
Application domain: General AI research
AI methods: Large language models
Research topics: eess.AS, cs.CL, cs.SD
arXiv record date: 2025-05-23
Publication date: 2025-05-23
Publication: Proc. Interspeech 2025, 3239-3243
DOI: 10.21437/Interspeech.2025-1292
Upstream last modified: Aug 26, 2025
Last record update: Nov 3, 2026

Abstract

The rapid growth of voice assistants powered by large language models (LLM) has highlighted a need for speech instruction data to train these systems. Despite the abundance of speech recognition data, there is a notable scarcity of speech instruction data, which is essential for fine-tuning models to understand and execute spoken commands. Generating high-quality synthetic speech requires a good text-to-speech (TTS) model, which may not be available to low resource languages. Our novel approach addresses this challenge by halting synthesis at the semantic representation level, bypassing the need for TTS. We achieve this by aligning synthetic semantic representations with the pre-trained Whisper encoder, enabling an LLM to be fine-tuned on text instructions while maintaining the ability to understand spoken instructions during inference. This simplified training process is a promising approach to building voice assistant for low-resource languages.